Improved ocean current measurements from gliders through in-situ compass calibration and improved flight models: dealing with strong gradients, shallow waters and no-surface dives

Abstract

Listen

Translate article

<p>Measuring ocean currents from gliders, either as dive-averaged currents or with acoustic doppler current profilers, requires accurate knowledge of the glider’s subsurface position and trajectory. These are estimated using dead-reckoning flight models which rely on accurate heading and speed estimates. We present two easily applicable and complementary methods to improve heading and speed estimates. These methods can be applied post-mission assuming adequate data were collected.</p> <p><span>Method 1:</span> Heading is determined using geomagnetic compasses on gliders and ADCPs. Compasses require calibration as they are affected both by the magnetic field of the glider itself and the electromagnetic field generated by sensors and motors. As of 2022, the majority of glider users rely on land-based, pre-deployment calibrations which are known to be unreliable due to user error, magnetic contamination from tools used in the calibration and changes in the glider’s magnetic field during transport, storage, power cycling and transport. These heading errors can be as high as 10 degrees. A simple correction scheme applied to compass magnetometer data improves heading estimates and current estimation for gliders and ADCP. Validation using a GPS compass shows a reduction in heading error to 2 degrees. A demonstration of the correction scheme and validation of current estimates is followed by recommendations for glider pilots and manufacturers to make these corrections possible for all glider deployments.</p> <p><span>Method 2:</span> Speed through water is determined using ocean glider flight models. These models rely on steady-state assumptions and decrease in accuracy in shallow waters, repeated no-surface dives or exceptionally strong density gradients. Accounting for acceleration and deceleration is possible but existing algorithms are computationally expensive. We propose a rapid and simple filtering approach which permits simultaneous representation of dynamic forces and empirical regression of flight parameters. We demonstrate improvements by comparing ADCP-derived velocities to dead-reckoning model velocities in the highly-stratified shallow waters of the Baltic where gliders surfaced on average once every 6 dives.</p>